fix(wecom): add CLOSING state to _wait_for_handshake terminal handler

[mydearzsy]

What

Adds aiohttp.WSMsgType.CLOSING to the terminal state check in _wait_for_handshake(), matching the identical fix already applied to _read_events() in #28311 (for issue #28293).

Why

_wait_for_handshake() was missing WSMsgType.CLOSING from its terminal state handler, causing WeCom reconnection to fail permanently with "closed during authentication" when the server sends a CLOSING frame during the authentication handshake.

When #28311 fixed _read_events() for issue #28293, _wait_for_handshake() was overlooked — both functions handle WebSocket messages and need to recognize all terminal states.

Impact

In production with multiple WeCom agents: after a mass restart event, agents that receive a CLOSING frame during the reconnection handshake get stuck in a permanent failure loop — manual pod restart is required to recover.

Fix

One-line change in plugins/platforms/wecom/adapter.py:

-            elif msg.type in {aiohttp.WSMsgType.CLOSED, aiohttp.WSMsgType.CLOSE, aiohttp.WSMsgType.ERROR}:
+            elif msg.type in {aiohttp.WSMsgType.CLOSED, aiohttp.WSMsgType.CLOSE, aiohttp.WSMsgType.ERROR, aiohttp.WSMsgType.CLOSING}:

Identical to the pattern already in _read_events() (line ~370).

Fixes: #64703

[tonydwb]

Code Review Summary

Verdict: Approved

Overview

Tiny fix: adds CLOSING state to the WeCom _wait_for_handshake terminal handler. 1-line addition.

Assessment

• Correctness: CLOSING state needed for proper terminal handler state machine coverage.
• Security: No security changes.
• Debug artifacts: None.

Summary

Trivially correct. LGTM.

---

Reviewed by Hermes Agent

[teknium1]

Thanks for identifying the unhandled handshake state. The premise holds on current main: plugins/platforms/wecom/adapter.py:332 omits WSMsgType.CLOSING, while the sibling receive loop handles it at plugins/platforms/wecom/adapter.py:373.

Problems

• The one-line behavior change has no regression test. tests/gateway/test_wecom.py does not exercise _wait_for_handshake() with a CLOSING frame.

Suggested changes

• Add an async fake-WebSocket test that returns aiohttp.WSMsgType.CLOSING and asserts the authentication-close RuntimeError.

The hunk remains unchanged since the PR base (6997dc81); salvaging this fix onto current main should otherwise be mechanical. This is an automated hermes-sweeper review.

[teknium1]

Please add a regression test with a fake WebSocket returning WSMsgType.CLOSING, asserting _wait_for_handshake() raises the authentication-close RuntimeError. Current tests/gateway/test_wecom.py has no coverage for this handshake state.

[GottZ]

This was generated by AI during triage.

Summary

Three PRs address #64703 by adding WSMsgType.CLOSING to _wait_for_handshake() so authentication-time closing frames raise immediately; #64716 additionally includes regression tests for CLOSING, CLOSED, and ERROR frames.

Related pull requests

• fix(wecom): add CLOSING state to _wait_for_handshake terminal handler #64707 [closed] duplicate — (+1/-1) — n/a: Adds the missing CLOSING terminal state and directly matches the reported root cause; although closed, it remains relevant as the first equivalent fix attempt, superseded by fix(wecom): add CLOSING state to _wait_for_handshake terminal handler #64716.
• fix(wecom): add CLOSING state to _wait_for_handshake terminal handler #64716 best fix — (+93/-1) — n/a: Adds the targeted CLOSING-state handler plus async regression tests for CLOSING, CLOSED, and ERROR frames. The current diff addresses the contributor keep_open review by including the requested fake-WebSocket CLOSING test that asserts the authentication-close RuntimeError.
• fix(wecom): add CLOSING state to _wait_for_handshake auth guard (#64703) #64727 [closed] duplicate — (+1/-1) — n/a: Applies the same one-line CLOSING-state fix and directly matches the reported omission; it remains relevant as a closed duplicate of fix(wecom): add CLOSING state to _wait_for_handshake terminal handler #64716, consistent with the contributor's canonicalization note.

Duplicates

#64707 and #64727 duplicate the production-code change in #64716; #64716 additionally carries regression coverage.

Suggested consolidation

Keep #64716 open with a salvage path: preserve its one-line root-cause fix and the regression tests that satisfy the contributor review, while rebasing them onto current main as needed. Keep #64707 and #64727 closed as duplicates of #64716.

Complex graph

flowchart LR
    classDef open fill:#dbeafe,stroke:#1d4ed8,color:#1e3a8a
    classDef merged fill:#dcfce7,stroke:#15803d,color:#14532d
    classDef closed fill:#e5e7eb,stroke:#6b7280,color:#1f2937
    classDef unverified fill:#f3f4f6,stroke:#9ca3af,color:#374151
    classDef best stroke-width:3px,stroke:#b45309
    classDef target stroke-width:3px,stroke:#4338ca
    I64703(["issue #64703 (open)"])
    subgraph Dup64707 ["PRs duplicating each other"]
        P64707["PR #64707 (closed)"]
        P64716["PR #64716 (open)"]
        P64727["PR #64727 (closed)"]
    end
    P64716 -->|best fix| I64703
    class I64703 open
    class P64707 closed
    class P64716 open
    class P64727 closed
    class P64716 best
    class P64716 target
    click I64703 "https://github.com/NousResearch/hermes-agent/issues/64703"
    click P64707 "https://github.com/NousResearch/hermes-agent/pull/64707"
    click P64716 "https://github.com/NousResearch/hermes-agent/pull/64716"
    click P64727 "https://github.com/NousResearch/hermes-agent/pull/64727"

Loading

Graph: solid arrow = fixes / best fix, dashed arrow = partial or unverified (see edge label); boxed group = PRs duplicating each other; amber border = best fix; indigo border = target; gray node = closed (state tag in the node label).

Cross-PR triage: Reviewed 3 pull requests and 1 issue in this complex. Each diff was read against this issue; Assessment working set: 6 kB of PR diffs, 7 kB of issue/PR text, 2 kB of discussion (4 comments), 4 verify verdicts. verdicts reflect diff content, not PR titles. Part of an automated triage batch.